|
helno posted:There is a battery technology designed just for you. They are called NIMH. 
|
#
?
Nov 1, 2015 16:37
|
|
|
|
| # ? Apr 27, 2024 14:52 |
|
|
In other news I finally got started on the aerodynamic Krieger knock off. Solidworks is a pain in the dick when your left mouse button doesn't work reliably. Got tired of waiting for the replacement micro switches so I just swapped the left and center switches for now. 
|
|
#
?
Nov 1, 2015 17:34
|
|
|
I just stumbled across a youtube video for some useful-looking software. https://www.youtube.com/watch?v=6R8WhMPl-54 I don't know how it works without a rangefinder mounted co-axially with the camera, but the idea is you send it up and it takes pictures of poo poo and turns it into a 3D model you can examine at leisure after your flight drone lands. This beats the pants off pausing/restarting 1 frame at a time to eke all the information possible out of each video taken. If this program can remember where it's been and allow you to stitch together data from multiple flights, it should be possible to map out an arbitrarily large area. In personal practical terms this means that I personally can not only map out where exactly all my trees are, but determine the optimal path through the woods to lay pex tube in the forest, and calculate exactly how much I need to buy. This is huge. But it's DJI only. It seems like something APM should be able to do though. Is there a menu dropdown I might have overlooked?
|
|
#
?
Nov 1, 2015 18:28
|
|
|
Good use for my dust-gathering P2V+, awesome I'll check it out.
|
|
#
?
Nov 1, 2015 20:35
|
|
|
DreadLlama posted:I just stumbled across a youtube video for some useful-looking software. It's not DJI only, the name should be a giveaway. Google 'orthomosaic mapping'.
|
|
#
?
Nov 2, 2015 00:52
|
|
|
After an afternoons work.
|
|
#
?
Nov 2, 2015 04:24
|
|
|
So I'm wiring up my quad, the ESC's are the DYS SN20A Opto. Do i need to wire up the ground signal lead, or is it safe to ignore it? I would imagine it could use the Power Ground connection as a reference?
|
|
#
?
Nov 2, 2015 10:37
|
|
|
Slash posted:So I'm wiring up my quad, the ESC's are the DYS SN20A Opto. Signal integrity is deeply related to path to ground. Also, the twisted wires shield the signal wire. It "might" work if you don't hook up the ground wire. But it also might not. Most likely, it'll work "most of the time" but you'll get random motor shutdowns in flight and be wondering why the hell your quad keeps doing wacked rear end poo poo in the air. Don't do that to yourself.
|
|
#
?
Nov 2, 2015 12:39
|
|
|
Nerobro posted:Signal integrity is deeply related to path to ground. Also, the twisted wires shield the signal wire. It "might" work if you don't hook up the ground wire. But it also might not. Most likely, it'll work "most of the time" but you'll get random motor shutdowns in flight and be wondering why the hell your quad keeps doing wacked rear end poo poo in the air. Ok will do. The PDB i'm using does not have pads connecting the Signal Ground to the flight controller, can i just solder it directly to the Ground connection on the PDB? 
|
|
#
?
Nov 2, 2015 13:38
|
|
|
Nerobro posted:Signal integrity is deeply related to path to ground. Also, the twisted wires shield the signal wire. It "might" work if you don't hook up the ground wire. But it also might not. Most likely, it'll work "most of the time" but you'll get random motor shutdowns in flight and be wondering why the hell your quad keeps doing wacked rear end poo poo in the air. While plenty have wired up their opto ESCs (including the sn20a) without the second ground fine, I'd leave it on just so that you can reprogram your ESCs whenever you want
|
|
#
?
Nov 2, 2015 14:22
|
|
|
PirateDentist posted:I dusted off my Nano QX since I really want to get into 250 racing quads next year, but I feel I need to get some solid hours in with this little guy before moving that direction. A good transmitter is a decent idea (look for a used dx7 or something) but the nano sucks in agility mode. It's too small and quick. Use self level. Also, I've found that some upgraded motors makes a big difference on it.
|
|
#
?
Nov 2, 2015 14:31
|
|
|
Slash posted:Ok will do. The PDB i'm using does not have pads connecting the Signal Ground to the flight controller, can i just solder it directly to the Ground connection on the PDB? Don't get me started on PDBs. :-) IF they want you to ignore the ground on the ESC, ignore the ground. Wojcigitty posted:It's too small and quick. Use self level. Mass makes such a difference in control ability. This is why bigger helicopters are easier to fly.
|
|
#
?
Nov 2, 2015 21:22
|
|
|
Eh, it's slowly coming together. Tuning the printer, printing the parts, fixing tolerances and reprinting, and the drat printer throwing curve balls, all at the same time, drags things out a little. :|  helno posted:After an afternoons work.
|
|
#
?
Nov 3, 2015 19:22
|
|
|
Combat Pretzel posted:Eh, it's slowly coming together. Tuning the printer, printing the parts, fixing tolerances and reprinting, and the drat printer throwing curve balls, all at the same time, drags things out a little. :|
|
|
#
?
Nov 4, 2015 15:37
|
|
|
Pix4d is really neat, the app is a little clunky but that might be because my p2v+ has a lot of hours on it (and at least one crash). Of course my GIS experience is lacking so what to do with these cool 3d models and maps is my next step.
|
|
#
?
Nov 4, 2015 15:48
|
|
|
BastionPete posted:This is real nice. How heavy are the 3D printed parts compared to, say, an aluminum set?  How stable it'll all be, I'll see on the first crashes. Those laser sintered Nylon sheets from Shapeways in my previous custom drone felt like Pringles chips, and were easy enough to break if you wanted to, yet it survived all crashes well enough considering. The test pieces I've printed in PETG feel way more durable.
|
|
#
?
Nov 4, 2015 17:29
|
|
|
So yeah, I had first contact with Arduinos and it went well. Now I have the stupid idea of a(nother) winter project of programming my own flight controller. I know of the APM 2.6, but it's the old 8-bit Arduino with a built-in IMU. Does anyone know of 32bit platforms like this, or am I out of luck and need to get a stock one and use a separate IMU?
|
|
#
?
Nov 4, 2015 19:37
|
|
|
Combat Pretzel posted:So yeah, I had first contact with Arduinos and it went well. Now I have the stupid idea of a(nother) winter project of programming my own flight controller. I know of the APM 2.6, but it's the old 8-bit Arduino with a built-in IMU. Does anyone know of 32bit platforms like this, or am I out of luck and need to get a stock one and use a separate IMU? Combat Pretzel posted:Density wise, aluminium is 2.7g/cm3 where as PETG (which I print in) is 1.4g/cm3. Plus, the insides are usually printed with sparse infill, making it a little lighter, like so: BastionPete fucked around with this message at 01:28 on Nov 5, 2015 |
|
#
?
Nov 5, 2015 01:23
|
|
|
As said, my Shapeways parts survived way more than I ever expected from them, and replacement parts I've printed myself are stronger. I have ABS filament here, but it's a pain in the rear end to print with. PET is what plastic bottles are made of, and you know how strong they are. Well, we'll see. A friend of mine had a custom design made from aluminium for a while. The arms were water jet cut from 5mm aluminium plates. They bent rather quickly on even lighter crashes. Not sure what alloy it was, tho. As far as a custom FC goes, quaternions is the only issue, never done these yet. A personal code base is always easier to work with. I want eventually things like smooth mode switchover (to mitigate jerking e.g. when switching from altitude hold to autolevel while throttle isn't in proper position) and, depending on how reliable I can measure it from the accelerometers and the barometer, throw-up-in-the-air arming (there's plenty of shrubbery around here for soft crashing while testing  ).I think I'll be going for the Arduino Zero instead, tho. It's smaller and I don't think that 8-bit PWM will pose a problem. Combat Pretzel fucked around with this message at 01:58 on Nov 5, 2015 |
|
#
?
Nov 5, 2015 01:56
|
|
|
Maybe it's just my ignorance, but surely a 3d printed part won't be the same strength as an identical part made using the same material, just with a more conventional manufacturing process? It looks like a nice little board. The DUE is faster, but I never got close to using it's full processing speed so that shouldn't be too much of a problem. For the PWM, remember the pulse needs to vary between 1 and 2 ms for the full range of motion/throttle. If you have a 10ms total pulse (100Hz), you will always be losing a lot of resolution on signal that doesn't matter. You have 256 different possible duty cycles on an 8bit pwm, but you're only interested in about 25 of them. This means only 25 unique throttle settings. Might not be a problem, I don't really know. Check it before you buy it. Quaternions are a bit of a bitch to get your head around. You'll also need to get to grips with axis-angle representations and euler angles. Find some libraries to convert between them. There are some pretty lovely resources out there on quaternions, some acceptable ones too. I looked at this well over a year ago now and can't remember any of it lol. To be honest, you don't have to fully understand what they are to use them effectively. So long as you understand their properties and can mindlessly use some stolen function to convert between quaternions, axis-angle and euler, you should be okay. The main thing to remember is that whenever you are calculating any kind of rotation, you want to be using quaternions so as not to get gimbal lock. When wanting to figure out thrust outputs it's useful to convert the calculated quaternion into a euler representation to apply it to the quadcopter. This probably won't make sense now. I'd recommend using madgwicks filter to get your orientation as a quaternion. Life is too short to write that poo poo yourself. Not as good as a kalman filter apparently, but acceptable for a multirotor. EDIT: poo poo. for the PWM you should be okay for the inputs as you can read it easily using interrupts and get full resolution on the pulse width. For the outputs I actually used adafruits I2C pwm driver because it was easier. EDIT2: Holy poo poo, my current MPU I'm working on only uses 8bit PWM outputs too... maaaaaybe need to think about this a bit more. EDIT3:well gently caress. I've been banging my head against a brick wall for ages trying to get the PWM working on my chip. The interrupts were just not playing ball and it was really slow for some reason. Still couldn't get it working last night. Now I get to throw it all away and go back to this. Well I mean the chip from the device and I'm planning on a custom PCB... EDIT4: gently caress everything. One of the reasons I chose my MCU in the first place is that it had a shitload of PWM outputs. Now I can't even use them. I even passed on a better chip (with an FPU) because it didn't have many PWM outputs. BastionPete fucked around with this message at 03:04 on Nov 5, 2015 |
|
#
?
Nov 5, 2015 02:24
|
|
|
All those edits aren't exactly making me hopeful. 
|
|
#
?
Nov 5, 2015 04:11
|
|
|
Combat Pretzel posted:Oooh nice. Update.  Those PET parts look good. Might have to try some different filament.
|
|
#
?
Nov 5, 2015 04:22
|
|
|
https://www.youtube.com/watch?v=HC89KRV7l9M
|
|
#
?
Nov 5, 2015 07:50
|
|
|
Combat Pretzel posted:All those edits aren't exactly making me hopeful.
|
|
#
?
Nov 5, 2015 12:05
|
|
|
Seems fake. As soon as the pumpkin hits, the camera is immediately on the ground and spinning, and then you can see a guy in the video at 30 seconds, but then apparently it took them 4 hours to find the camera. Not sure why though, because the actual shot of the pumpkin seems real, as if it barely missed the drone but they decided to make it look like it hit instead? I think a shot of a pumpkin from a cannon whizzing by would be just as impressive.
|
|
#
?
Nov 5, 2015 14:26
|
|
|
Erwin posted:Seems fake. As soon as the pumpkin hits, the camera is immediately on the ground and spinning, and then you can see a guy in the video at 30 seconds, but then apparently it took them 4 hours to find the camera. Not sure why though, because the actual shot of the pumpkin seems real, as if it barely missed the drone but they decided to make it look like it hit instead? I think a shot of a pumpkin from a cannon whizzing by would be just as impressive. That's a lot of words man. Looks real to me.
|
|
#
?
Nov 5, 2015 16:23
|
|
|
dev null posted:That's a lot of words man. Looks real to me. Same, except it took 4 hours to find the thing? That was what I found weird I saw the person in the video too, even managed to find a still frame:  It's some slenderman type poo poo. Like the last frame captured on a camera recovered from a dead body. 
|
|
#
?
Nov 5, 2015 16:45
|
|
|
A photo of the person who threw the gopro with a wicked topspin maybe
|
|
#
?
Nov 5, 2015 17:28
|
|
|
HOW CREEPY IS THAT
|
|
#
?
Nov 5, 2015 17:29
|
|
|
Erwin posted:Seems fake. As soon as the pumpkin hits, the camera is immediately on the ground and spinning, and then you can see a guy in the video at 30 seconds, but then apparently it took them 4 hours to find the camera. Not sure why though, because the actual shot of the pumpkin seems real, as if it barely missed the drone but they decided to make it look like it hit instead? I think a shot of a pumpkin from a cannon whizzing by would be just as impressive. Pumpkin cannons exist. Camera drones exist. Is it really that hard to believe that a 200mph+ flying pumpkin could veer slightly off course and hit the drone? Go to the youtube link, the guy's posted pictures of the damage. The camera spins while it's falling, not when it's on the ground, and I can tell you from experience that finding something lost in the woods is way more difficult than it seems, even if "I swear I just saw it land right here."
|
|
#
?
Nov 5, 2015 19:53
|
|
|
Sagebrush posted:Pumpkin cannons exist. Camera drones exist. Is it really that hard to believe that a 200mph+ flying pumpkin could veer slightly off course and hit the drone? Go to the youtube link, the guy's posted pictures of the damage. Address the blurred X-Files-intro-esque abomination! VV EDIT: The guy is pictured WHILE it's spinning. Just before it finally stops. It spun for 4 hours, cut for time as you said. The appearance of the man disrupted whatever wormhole the gopro got caught in. Probably(obviously) playful Greys. bring back old gbs fucked around with this message at 20:04 on Nov 5, 2015 |
|
#
?
Nov 5, 2015 19:58
|
|
|
yeah well uh how come there isnt 4 hours of sitting in the grass footage since they said it took 4 hours to find so uh are they lying now or were they lying then
|
|
#
?
Nov 5, 2015 20:01
|
|
|
PUMPKINS CANT MELT STEEL BEAMS
|
|
#
?
Nov 5, 2015 20:15
|
|
|
What I find interesting isn't that it recorded nothing. Its that it recorded 4 hours of nothing Contact still haunts me.
|
|
#
?
Nov 5, 2015 21:12
|
|
|
https://www.youtube.com/watch?v=mL0iC0HPbhg DJI's first combat loss? I also almost had one of my own, during more... vigorous banking, the balancing cable slipped out from wherever I tucked it in, and got hit by a prop, cutting one wire clean off and loving up the connector. Thankfully it didn't short out while in the air so hopefully cannibalizing a connector off a different battery would be enough to get up in the air again.  I cut the yellow one myself before finding the cork in my glovebox, thankfully I keep all sort of junk in the car.
|
|
#
?
Nov 6, 2015 20:04
|
|
|
Anyone have any idea what it would take to haul a LIDAR unit into the air and do surveys?
|
|
#
?
Nov 6, 2015 21:26
|
|
|
Elendil004 posted:Anyone have any idea what it would take to haul a LIDAR unit into the air and do surveys? That's a pretty vague question. It would take a big airplane and an onboard computer and some way to fly it on a specified path.
|
|
#
?
Nov 7, 2015 00:36
|
|
|
porksmash posted:That's a pretty vague question. It would take a big airplane and an onboard computer and some way to fly it on a specified path. You can get pretty small LIDAR units now.
|
|
#
?
Nov 7, 2015 01:14
|
|
|
porksmash posted:That's a pretty vague question. It would take a big airplane and an onboard computer and some way to fly it on a specified path. Obviously on an RC aerial vehicle.
|
|
#
?
Nov 7, 2015 04:57
|
|
|
|
| # ? Apr 27, 2024 14:52 |
|
|
Elendil004 posted:Anyone have any idea what it would take to haul a LIDAR unit into the air and do surveys? http://pulsedlight3d.com/blogs/cool-stuff/25137473-pixhawk-lidar-lite-setup-demo
|
|
#
?
Nov 7, 2015 05:43
|
|